AI Is a Tool. Using It Well Takes Skill.

jgord warned that AI may roll out faster than humans can learn to use it responsibly, predicting a decades-long drop in cognitive ability and education; OptionOfT described immediate job-market pain and uncertainty; sriram_malhar insisted we call out "AI companies" for harvesting creative work and concentrating wealth; im_down_w_otp compared current models to a slot machine rather than a predictable tool; deadbabe called AI an "anti-tool" that uses people to get work done; and beloсh pointed to incidents like the HuggingFace hack as evidence that even builders sometimes lack control. Several commenters (whatever1, schleck8) argued that model progress is already making previously valuable prompting or tooling skills obsolete, and BatchJob and strangescript noted that nondeterminism and persistent errors mean outputs still require human correction.

Others defended the "AI-as-tool" framing or offered nuance: mosesman and gerdesj compared AI to domain expertise in photography or a helpful search assistant, arguing experience still matters; vanuatu and girvo listed concrete skills - verification, guardrails, context engineering, model selection, infrastructure - that they say remain necessary; jonstaab suggested both views are true, that tools instrumentally shape us while also reshaping our capacities; ProblemExplorer and others framed advanced agents as new categories requiring different skills. The split centers on whether human skill will remain essential as models improve, and whether AI should be treated as a controllable tool, a collaborator with agency, or an industry shaped by corporate incentives.